veekaybee

Anti-hype LLM reading list

Goals: Add links that are reasonable and good explanations of how stuff works. No hype and no vendor content if possible. Practical first-hand accounts of models in prod eagerly sought.

Foundational Concepts

Pre-Transformer Models

lotharschulz

Publishing artifacts with AWS Codeartifact and GitHub Packages

• Blog Post: Publishing artifacts with AWS Codeartifact and GitHub Packages
• GitHub repository with full code details: reach-now/codeartifact-packages-publishing

AWS Codeartifact

Repository Setup

aws cli

djg07

#Talk python to me

import json
import boto3

s3 = boto3.client('s3')

def lambda_handler(event, context):
   bucket = 'aws-simplified-transactions'
   key = 'transactions.json'

moxious

#!/bin/bash
#
# This startup script replaces the normal neo4j startup process for cloud environments.
# The purpose of the script is to gather machine IP and other settings, such as key/value
# pairs from the instance tags, and use that to configure neo4j.conf.
#
# In this way, neo4j does not need to know ahead of time what it's IP will be, and
# can be controlled by tags put on the instance.
######################################################################################
echo "pre-neo4j.sh: Fetching AWS instance metadata"

bendog

AWS Lambda JWT authenticaiton

AWS API Gateway has the ability to pre-authenticate connections prior to launching the endpoint, by passing the authorizationToken to a Lambda function. There are clear benefits for simplifying end point security and also a reduction in duplicated code by utilising this feature. However I found the AWS examples were excessively complicated for what should be a very simple task.
So here's my example.

The main concern is that AWS Lambda authentication expects a very specific response and if that response is not given it will throw a 500 error with x-amzn-ErrorType: AuthorizerConfigurationException in the response header if the response object is not exactly as expected.

Usage

I personally use to handle the publishing part of my Lambdas, but I'll include an image of the API Gateway config.

gwhitelaw

This is how I quickly got an Apache Zepplin notebook running against the AWS Glue Dev endpoint. None of the guides out there seemed concise, and I found some custom Docker containers doing what you can do easily. This gives you the power - it sets up port forwarding & runs the official Docker image.

1. Create your Glue Dev endpoint (this involves creating a keypair, I just used ssh-keygen)
2. Once READY, select it and copy the "SSH tunnel to remote interpreter"

• eg: ssh -i <private-key.pem> -vnNT -L :9007:169.254.76.1:9007 glue@..compute.amazonaws.com

3. Connect to the endpoint in a terminal session, modifying the above to match: ssh -i ~/.ssh/glue-dev -vnNT -L :9007:*127.0.0.1*:9007 glue@<ec2-endpoint>.<region>.compute.amazonaws.com
4. Run the Apache Zepplin Docker container `docker run -p 8080:8080 --rm -v $PWD/logs:/logs -v $PWD/notebook:/notebook -e ZEPPELIN_LOG_DIR='/l

rbalicki2

How we incorporate next and cloudfront (2018-04-21)

Feel free to contact me at robert.balicki@gmail.com or tweet at me @statisticsftw

This is a rough outline of how we utilize next.js and S3/Cloudfront. Hope it helps!

It assumes some knowledge of AWS.

Goals

shortjared

a4b.amazonaws.com
access-analyzer.amazonaws.com
account.amazonaws.com
acm-pca.amazonaws.com
acm.amazonaws.com
airflow-env.amazonaws.com
airflow.amazonaws.com
alexa-appkit.amazon.com
alexa-connectedhome.amazon.com
amazonmq.amazonaws.com

mojodna

--
-- This will register the "planet" table within your AWS account
--
CREATE EXTERNAL TABLE planet (
  id BIGINT,
  type STRING,
  tags MAP<STRING,STRING>,
  lat DECIMAL(9,7),
  lon DECIMAL(10,7),
  nds ARRAY<STRUCT<ref: BIGINT>>,

bhtucker

"""
Upsert gist

Requires at least postgres 9.5 and sqlalchemy 1.1

Initial state:

[]
Initial upsert: